| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.444444444444445, | |
| "eval_steps": 500, | |
| "global_step": 30, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.14814814814814814, | |
| "grad_norm": 14.376841688418876, | |
| "learning_rate": 4.986304738420684e-06, | |
| "loss": 0.9343, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.7407407407407407, | |
| "grad_norm": 17.11675159291135, | |
| "learning_rate": 4.665063509461098e-06, | |
| "loss": 0.5078, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 1.4814814814814814, | |
| "grad_norm": 1.9614630752880515, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 0.2162, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 1.2395316810250545, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.1698, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 2.962962962962963, | |
| "grad_norm": 1.1880248562815259, | |
| "learning_rate": 1.2500000000000007e-06, | |
| "loss": 0.1391, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 3.7037037037037037, | |
| "grad_norm": 0.790459548974325, | |
| "learning_rate": 3.3493649053890325e-07, | |
| "loss": 0.1234, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 4.444444444444445, | |
| "grad_norm": 0.678402289189651, | |
| "learning_rate": 0.0, | |
| "loss": 0.1205, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 4.444444444444445, | |
| "step": 30, | |
| "total_flos": 4291380330496.0, | |
| "train_loss": 0.2270124097665151, | |
| "train_runtime": 648.6643, | |
| "train_samples_per_second": 3.33, | |
| "train_steps_per_second": 0.046 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 30, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4291380330496.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |